Passing the Buck to the Wealthier: Reference-Dependent Standards of ☆ T Generosity ⁎ Jonathan Z

Organizational Behavior and Human Decision Processes 157 (2020) 46–56

Contents lists available at ScienceDirect

Organizational Behavior and Human Decision Processes

journal homepage: www.elsevier.com/locate/obhdp

Passing the buck to the wealthier: Reference-dependent standards of ☆ T generosity ⁎ Jonathan Z. Bermana, , Amit Bhattacharjeeb, Deborah A. Smallc, Gal Zaubermand a Associate Professor of Marketing at the London Business School, United Kingdom b Associate Professor of Marketing at INSEAD, France c Laura and John J. Pomerantz Professor of Marketing and Psychology at the University of Pennsylvania, United States d Joseph F. Cullman, 3rd Professor of Marketing at Yale School of Management, United States

ARTICLE INFO ABSTRACT

Keywords: Who is expected to donate to charity, and how much should they give? Intuitively, the less financially con- Charitable giving strained someone is the more they should give. How then do people evaluate who is constrained and who has Spare money money to spare? We argue that perceptions of spare money are reference-dependent with respect to one’s current Resource slack self: those who earn more than oneself are perceived as having an abundance of spare money and thus as Financial decision making ethically obligated to donate. However, those higher earners themselves report having little to spare, and thus apply lower donation standards to themselves. Moreover, a meta-analysis of our file-drawer reveals an asymmetry: individuals overestimate the spare money of higher earners but estimate the scant spare money of lower earners more accurately. Across all incomes assessed, people “pass the buck” to wealthier others (or to their future wealthier selves), who in turn, “pass the buck” to even wealthier others.

1. Introduction person. Poor people who are already struggling to meet their own basic needs cannot help others without substantial self-sacrifice. Meanwhile, How much money do people think they and others should donate to the rich can easily afford to give to others without sacrificing nearlyas charity? Existing research typically examines how various contextual much. factors can motivate or discourage giving in specific, isolated instances However, subjective assessments of financial well-being often differ (for summaries, see Bekkers & Wiepking, 2011; Slovic, 2007; Small & markedly from objective reality. One survey found that only 31% of Cryder, 2016). In contrast, little research has examined judgments American millionaires consider themselves to be wealthy (UBS Investor about how much different individuals ought to sacrifice in order tohelp Watch, 2013). Other surveys have found similar results for even richer others (cf. LaBarge & Stintson, 2014; Piff, Kraus, Côté, Cheng, & individuals: only 36% of those with $5 million in assets consider Keltner, 2010; Sussman, Sharma, & Alter, 2015). themselves “financially secure” (Schervish & Havens, 2001), and only This paper investigates individuals’ subjective ethical standards of 21% of households with $50 million in assets consider themselves to be giving for oneself and others across income levels. We expect that “extremely financially secure” (Rooney & Frederick, 2007). In contrast, people generally agree that the more one earns, the more money they ethicists have argued that even ordinary individuals in the developed can spare, and therefore the more they ought to donate to charity. This world are quite wealthy by global standards, and thus able to donate a intuition is consistent with the basic principle of diminishing marginal sizable portion of their income without significant self-sacrifice (Singer, utility from income: gaining (or giving up) an additional dollar matters 1979, 2009, 2015; MacAskill, 2015). For instance, an individual very little to a rich person, but can dramatically impact the life of a poor earning $50,000 in the US ranks within the top 1.5% of income earners

☆ The authors would like to thank Raghu Iyengar, Chris Olivola, Stephen Spiller, Shereen Chaudhry and seminar participants at Columbia University, Cambridge University, University of Michigan, Washington University—St Louis, University of Southern California, HEC Paris, Cass Business School, the MoRL lab group, and the University of Utah for their feedback on previous versions of this manuscript. ⁎ Corresponding author. E-mail addresses: [email protected] (J.Z. Berman), [email protected] (A. Bhattacharjee), [email protected] (D.A. Small), [email protected] (G. Zauberman). https://doi.org/10.1016/j.obhdp.2019.12.005 Received 27 November 2018; Received in revised form 9 October 2019; Accepted 20 December 2019 0749-5978/ © 2019 Elsevier Inc. All rights reserved. J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56 worldwide, even after accounting for differences in purchasing power.1 richer others in a similar fashion. Hence, people across income levels In this paper, we examine how an individual’s own income affects may maintain the belief that they are justified in giving little because the subjective standards of generosity they apply to themselves and to they have little to spare, while believing that higher earners could ea- others. Specifically, we ask people to report how much they believe sily donate the money they spend in seemingly extravagant ways. In someone with a specified income should donate to charity each year, other words, individuals across incomes feel obligated to donate sig- and find strong evidence that these evaluations are reference-dependent nificantly less money than poorer individuals believe they oughtto with respect to the self. Subjective ethical standards of giving not only give. depend on the target’s absolute income, but also on the evaluator’s own Over and above how people construe expenditures, they may also income, and, specifically, on how much more or less the evaluator earns misjudge how earning more income will affect their overall spending relative to that target. While the principle of diminishing marginal levels. People often fail to anticipate that they will succumb to temp- utility is consistent with the basic notion that higher earners should tations to spend (Loewenstein, 1996; Peetz & Buehler, 2009) or incur give more than lower earners (because it is less painful for them to do unexpected expenses in the future (Sussman and Alter, 2012). Even so), it in no way suggests that people’s judgments of how much is ap- when people believe that their expenses will increase, they often un- propriate to give at each income level should depend on their own fi- derestimate the extent to which these increased expenses will deplete nancial situation. We find that, holding a target’s income constant, the their spare money (Berman, Tran, Lynch, & Zauberman, 2016). This less an evaluator earns relative to that target, the more money they tendency to under-account for increased future expenses appears to be believe that target ought to donate to charity. driven by an inability to properly anticipate the tradeoffs associated We further find that subjective giving standards are driven inpart with budgeting. For instance, “tightwad” individuals who naturally by erroneous perceptions of how earning more income generates ad- attend to opportunity costs do a better job of accounting for their future ditional spare money, or financial slack (Zauberman & Lynch, 2005). expenses than “spendthrifts” who are less attentive to opportunity costs Though people often feel financially constrained and report having (Berman et al., 2016; Frederick, Novemsky, Wang, Dhar, & Nowlis, little spare money in the present, they expect that earning a higher 2009). And those who actively calculate future tradeoffs by planning income will considerably ease their financial constraints, creating an for the long term have better credit ratings than those who rely on abundance of spare money. Accordingly, they believe that higher- intuitions to guide their finances (Lynch, Netemeyer, Spiller, & Zammit, earning others can easily donate to charity without experiencing sig- 2009). nificant self-sacrifice. In contrast, higher earners themselves report Together, these findings suggest that people will systematically having much less spare money than lower earners expect, and conse- overestimate the relationship between higher income (relative to one’s quently, believe they should donate much less. own income) and the availability of spare money, expecting higher In the following section, we further explain why an individual’s own earners to have more spare money than they actually report having. income is likely to affect how they perceive others’ finances, andwhy Whereas past research shows that people erroneously overestimate the these perceptions might affect the standards of charitable giving they extent to which increases in income will make them happier apply to others. (Kahneman et al., 2006), we predict that people will also erroneously overestimate the extent to which increases in income will increase the spare money available to them. 2. Income, spare money & subjective standards of giving We additionally expect that these misperceptions of spare money will increase with bigger income differences between oneself and a Past research points to a number of reasons why people’s own in- target individual. Thus, the less someone earns relative to a target, the come might influence how they evaluate the standards of giving applied more they will overestimate the spare money that target has available, to others across the income distribution. For one, people might fail to and the more they will believe that target is ethically obligated to do- appreciate that they and others hedonically adapt to changes in income nate to charity. (Frederick & Loewenstein, 1999; Kahneman, Krueger, Schkade, Lastly, in addition to overestimating the spare money of others with Schwarz, & Stone, 2006). The notion that increases in income and other a higher income, we expect these results to hold even when individuals positive life circumstances tend to produce only short-term gains in imagine how their own spare money would change if their income in- well-being is referred to as the “hedonic treadmill” (Brickman & creased. That is, we expect that people will overestimate the relation- Campbell, 1971; Deiner, Lucas & Scollon, 2006). Earning more money ship between increases in income and spare money regardless of whe- allows people to spend more on goods and services, and these addi- ther they evaluate themselves in the future or others who earn more tional expenditures are initially pleasurable. However, people often than they do in the present. adjust quickly to this increased standard of living. Not only does the The above hypotheses focus on how individuals make upward fi- pleasure from making these new purchases soon fade, but forgoing nancial evaluations, or assess the spare money and giving standards of these new expenditures is soon reconstrued as a painful loss as well higher-earning individuals. However, our predictions are less clear re- (Kahneman, Knetsch, & Thaler, 1991). garding whether individuals will over- or under-estimate the spare Failing to account for adjustment to living standards as incomes money of lower earners, and the nature of downward evaluations is grow may distort how people assess others’ available financial re- thus an empirical question. On the one hand, people may fail to ap- sources. For most people, charitable donations are discretionary ex- preciate that lower earners are accustomed to spending less and un- penses that are considered only after their own essential needs have derestimate how much spare money they have. On the other hand, it is been met. A poorer individual, struggling to make ends meet, might feel also possible that people fail to realize that lower earners also succumb little ethical obligation to donate (cf. Sharma, Mazar, Alter, & Ariely, to temptations to spend, and thus believe they have more spare money 2014), but is likely to characterize a richer person as spending a great than they actually do. deal of their income on nonessentials that could otherwise easily be In our studies, we find that individuals slightly underestimate the donated to charity. In contrast, a richer person who is accustomed to spare money of lower-earning others. Moreover, we find evidence of an their lifestyle may construe the same expenditures as necessities that asymmetry in upwards versus downwards evaluations of others: a meta- are far too painful to forgo. That richer person may, in turn, regard even analysis of our entire file-drawer shows that relative income differences are more pronounced in judgments of individuals who earn more than 1 See https://www.givingwhatwecan.org/get-involved/how-rich-am-i/ and oneself than in judgments of individuals who earn less than oneself. In https://politicalcalculations.blogspot.com/2016/10/what-is-your-world- other words, individuals overestimate the spare financial resources of income-percentile.html higher earners to a greater extent than they underestimate the spare

47 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56 financial resources of lower earners. Importantly, these judgments the more they would believe that earning this amount affords someone produce a corresponding asymmetry in judged standards of charitable greater spare money, and the more they believe that someone earning giving. this amount should donate to charity. In sum, we find that across all the income levels assessed, individuals believe that they themselves, as well as lower earners, are 4.1. Study 1a - method largely excused from obligations to donate to charity, but that higher earners should donate significantly more than higher earners them- Participants (Mechanical Turk, N = 505; Mean age = 32.5, 39% selves feel they should. Thus, regardless of how much they currently female; 61% male; Median Household Income = $45,000) were asked earn, individuals consistently “pass the buck” to higher-earning others to imagine what their life would be like if their combined household (or their own higher-earning future selves), who in turn “pass the buck” income was $50,000 per year. They then were asked to consider “what to even higher-earning others. you feel would be the appropriate amount of money you think that you should donate to charity if you earned this amount. This number should 3. Overview of studies reflect what you think is morally appropriate given that your household income is $50,000 per year.” Responses were given in dollars, which we We test these predictions in four studies and a meta-analysis of our log transformed (ln + 1) to account for skewness in the data. file-drawer. Study 1a investigates whether individuals earning less than Participants then evaluated the likelihood that they would be able to $50,000 believe that they would have much more spare money and make a one-time payment of $2000 without having to dip into their therefore should donate more to charity than what those currently retirement fund, borrow money, or charge it to a credit card, if their earning this target amount evaluate for themselves. Study 1b then in- household income was $50,000 a year (on a scale ranging from 1 = vestigates whether these judgments hold when making evaluations of “extremely unlikely” to 11 = “extremely likely”). This served as our others rather than an imagined future self. Study 2 examines whether measure of estimated spare money (Berman et al., 2016; Lusardi, these effects are robust across a wide range of target incomes. Study3 Schneider, & Tufano, 2011). further tests the psychological process, examining whether subjective At the end of the survey, participants reported how much they do- ethical standards of giving are directly influenced by perceptions of nated to charity in the previous year in dollars (Median = $100; spare money itself, above and beyond differences in income. Finally, we Mean = $539, SD = $1677), and estimated their own ability to make a present a meta-analysis of our file drawer (15 studies, N = 7163) to one-time payment of $2000 on a scale ranging from 1 = extremely investigate the robustness of these effects and examine whether eva- unlikely to 11 = extremely likely (M = 5.41, SD = 3.83). luations of higher earners differ systematically from evaluations of lower earners. 4.2. Study 1a - results In all studies, sample sizes were determined in advance. No conditions or participants were dropped from any analysis performed. All Subjective Giving Standards: A univariate linear regression with measures assessed that are not reported can be found in the supporting participant income predicting subjective giving standards shows that materials. Data files can be found online at(https://osf.io/s5vdf/). the less a respondent earns, the more they believe they should donate if they earned $50,000 a year (β = −0.19, SE = 0.03), t(503) = −5.70, 4. Study 1A & 1B p < .001. Fig. 1 displays these results. One may wonder if the reason why poorer people expect to donate In Study 1a, participants were asked to imagine what their lives more than richer people is that they currently donate a greater pro- would be like if they earned $50,000 a year, while Study 1b asks par- portion of their income to charity, and expect that they would donate ticipants to imagine a similar other earning $50,000 a year. We ex- the same proportion of their income if they earned $50,000 a year. To pected that the less that participants currently earn, relative to $50,000, address this possibility, we examined if our findings can be explained

Study 1a: How much money do you believe would be morally Fig. 1. Relationship between participants’ income and their judgments of how much they appropriate to donate to charity if you earned $50,000 per year? believe they should donate to charity with a $1,400 household income of $50,000 per year. Blue bars correspond to those who earn more than $50,000 and red bars correspond to those who $1,200 earn less than $50,000. Mean values and 95% confidence intervals for each income bracket are $1,000 computed on the log transformed dollar amounts $50,000 (ln + 1), are then transformed back into dollar $800 amounts for ease of interpretation. (For interpretation of the references to colour in this figure $600 legend, the reader is referred to the web version of this article.) $400

$200

Participants' Current Household Income

48 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56

Table 1 Robustness Checks, Study 1a.

Ln (Giving Standards + 1) Spare Money

I II III IV V VI

Intercept 6.31*** 6.05*** 4.69*** 5.06*** 8.68*** 7.62*** (0.21) (0.21) (0.66) (0.65) (0.25) (0.83) Income ($10,000 s) −0.19*** −0.20*** −0.23*** −0.22*** −0.37*** −0.36*** (0.03) (0.03) (0.04) (0.04) (0.04) (0.05) Age −0.02 −0.01 0.02 (0.01) (0.01) (0.02) Gender −0.06 0.04 −0.58† (0 = Male, 1 = Female) (0.24) (0.24) (0.30) Marital Status −0.06 −0.09 −0.49 (0 = Single, 1 = Married) (0.29) (0.29) (0.36) Number of children 0.11 0.16 −0.19 (0.13) (0.13) (0.16) Education 0.53*** 0.43* 0.24 (0.18) (0.18) (0.23) Political conservatism −0.11 −0.15† 0.01 (0.08) (0.08) (0.10) Religiosity 0.25*** 0.22** 0.03 (0.07) (0.07) (0.09) Tithing 1.29** 0.51 1.03 (0 = No, 1 = Yes) (0.45) (0.48) (0.61) Participants' Donation Rate (% of household income) 28.88*** 21.89*** −9.07 (4.17) (4.75) (6.00)

R-Squared 0.06 0.14 0.15 0.19 0.15 0.17

Regression coefficient and standard errors. The relationship between participant income (row 2) and both subjective giving standards and expected sparemoneyare robust to a wide range of control variables, including past self-report donation rates. †p < .10, * p < .05, ** p < .01, *** p < .001. by differential donation rates (as a percentage of income) among the Estimated Spare Money: We repeated the above univariate linear rich versus the poor. On average, our participants reported donating regression analysis with spare money as the dependent variable. Results 0.93% (SD = 2.6%) of their income to charity in the past year, and self- show that a participant’s current income negatively predicted their reported donations to charity were not significantly correlated with estimated ability to make a one-time payment of $2000 if they earned income (r = 0.05, p = .26). Controlling for individual donation rates $50,000 a year (β = −0.37, SE = 0.04), t(504) = −9.31, p < .001. does not affect the relationship between participant income and sub- Fig. 2 displays these results, and Table 1 presents results controlling for jective giving standards (β = −0.20, SE = 0.03), t(502) = −6.13, demographic variables. p < .001. Table 1 presents these results in full, along with demo- Thus, the less participants earn relative to $50,000, the more spare graphic covariates. In subsequent studies, all of our results remain money they expect to have and the more they believe they should do- highly robust to the inclusion of demographic covariates, including past nate to charity if they earned this amount. donation rates. See supporting materials for details.

Study 1a: Participants' Estimated Likelihood of Making Fig. 2. Relationship between participants’ income and their judgments of how much spare a $2,000 Emergency Payment if they Earned $50,000 a year money they would have if they earned $50,000. 11 = 11 Blue bars correspond to participants earn more Extremely 10 $50,000 than the target income ($50,000) and red bars likely correspond to those who earn less than the 9 target. Error bars correspond to 95% confidence intervals. (For interpretation of the references to 8 colour in this figure legend, the reader is referred 7 to the web version of this article.) 6 5 4 3 1 = 2 Extremely unlikely 1

Participants' Current Household Income

49 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56

4.3. Study 1b - methods As in Study 1b, participants in this study first reported their age, gender (80% female; 20% male), marital status (58% single, 42% In Study 1b (Mechanical Turk, N = 1002; Median Household married), and number of children (M = 1.59, SD = 1.42), were again to Income = $45,000), we test whether this relationship holds when imagine a target individual with demographic characteristics similar to people evaluate others. At the beginning of the survey, participants their own. However, in this study, we presented each participant with reported their age (M = 34.2, SD = 11.2), gender (45% female), mar- five such target individuals on separate pages, each with a different ital status (33% single, 67% married), and number of children income level: $20,000, $40,000, $60,000, $80,000, or $100,000 per (M = 0.74, SD = 1.21). Participants then imagined a target individual year. For each of the five targets, participants reported the percentage who had similar demographic characteristics as themselves. For ex- of household income that individual should donate to charity each year, ample, a participant who reported that he is a single male, 36 years old, which served as our measure of subjective ethical standards of chari- and has no children, was asked to imagine a target that is a 36-year-old table giving. The presentation order of the target individuals was dis- single male with no children. played in either ascending order (lowest to highest income) or des- All participants were told that the target individual’s household cending order (highest or lowest income) and counterbalanced across income was $50,000 per year. They then provided evaluations of how participants. Participants then repeated this process for our measure of much they believed that this person should donate to charity each year, spare money, again evaluating the five target individuals on separate which was measured as a percentage of total income in this study (as pages in the same order as the subjective giving standards measure they opposed to the dollar amount measure used in Study 1a). Participants completed. also evaluated the target’s estimated spare money in the same manner as Study 1a. 6.2. Results 4.4. Study 1b - results Subjective Giving Standards: To examine how a participant’s income Consistent with our previous results, participants’ current income relative to the target predicts subjective giving standards, we first cre- was a significant negative predictor of subjective giving standards ated a measure of relative income using the difference between how (β = −0.30, SE = 0.10, t(1000) = −2.93, p = .003), and estimated much each participant, i, earns relative to each target individual, j. spare money (β = −0.34, SE = 0.03, t(1000) = −12.41, p < .001). Fig. 3 plots the relationship between a participant’s relative income to each target and how much they believe that a given target should do- 5. Discussion nate to charity. We then conducted the following mixed model linear analysis, in- Together, Study 1a and 1b show that an individual’s household cluding Participant ID as a random effect, Relative Income (centered at income predicts how they evaluate the finances of those earning 0) as a continuous fixed effect, and Target Income and Presentation $50,000 a year, whether evaluating themselves or others. The less Order as categorical fixed effects. someone presently earns, the more they believe that earning $50,000 results in additional spare money, and the more they believe that a Giving Standardsij = ParticipantIDi + RelativeIncomeij + TargetIncomej person earning $50,000 should donate to charity. In Study 2 we test the + (RelativeIncomeij X TargetIncomej )+ PresentationOrderi robustness of these results by examining a wider range of target in- + ij (1) comes. We estimated the model in Eq. (1) and evaluated the simple effect of 6. Study 2 relative income at each level of target income. Table 2 displays these results. In Study 2, we make a number of changes to test the robustness of The coefficient for relative income is significant for each targetin- our findings. As in Study 1b, we again ask participants to evaluate come evaluated: a participant’s current income consistently affects how giving standards and spare money of a similar other. However, in the much they believe each target should donate to charity. Moreover, the present study, we ask participants to assess multiple target individuals coefficient is negative across the entire range of target individuals: the with different incomes in order to see whether our results hold acrossa less participants earn relative to a target, the more they believe that range of target income levels. In addition, we utilize a stratified sam- target should donate to charity, and the more participants earn relative pling plan to obtain enough participants across within each income to a target, the less they believe that target should donate. Together, level bracket assessed. Consistent with Study 1a & 1b, we expect that these results confirm our pre-registered hypothesis. the less participants earn relative to each target individual, the more Note that the effect size of relative income on subjective giving spare money they will believe the target has, and the more they will standards is larger when evaluating high-income targets (e.g., those believe the target should feel obliged to donate to charity. This study earning $100,000) compared to low-income targets (e.g., those earning design and analysis plan was pre-registered with the OSF (https://osf. $20,000). This suggests that individuals are particularly sensitive to io/2hfjk/). relative income when making upward evaluations of higher earners compared to when making downward comparisons of lower earners. 6.1. Methods We investigate this asymmetry further in the meta-analysis reported below. We recruited 1022 participants via an online panel. Specifically, we Estimated Spare Money: We computed the same model listed in Eq. sampled an equal number of participants in each of the following in- (1), replacing the DV with the measure of estimated spare money. Re- come brackets: $0 to $10,000; $10,000 to $30,000; $30,000 to sults are displayed in Table 2. Across all target incomes, the effect of $50,000; $50,000 to $70,000; $70,000 to $90,000; $90,000 to relative income is significant and negative: the less (more) that parti- $110,000; over $110,000. Within each income bracket, we sampled an cipants earn relative to a target, the more (less) likely they are to be- equal number of participants in two age groups, one between the ages lieve that target individual is capable of making a one-time payment of 18 to 40 (Mean Age = 30) and the other between the ages 41 to 65 $2,000. These results again support our pre-registered hypothesis. (Mean Age = 52).

50 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56

Subjective Ethical Standards of Giving for Each Target Fig. 3. Model free means relating how much participants believe that others should donate to Individual by Participants Relative Income Level charity as a function of their relative income to the target individual. The bars identify partici- Participant income relative to target pants’ income relative to each of the five target 50% $90 - $100k less individuals. Subjective giving standards are $70 - $90k less measured as a percent of target’s income. For 45% $50 - $70k less presentation purposes, relative income is $30 - $50k less grouped in $20,000 increments. 40% $10 - $30k less $10k less to $10k more 35% $10 - $30k more $30 - $50k more $50 - $70k more 30% $70 - $90k more $90k - $110k more 25%

20%

15%

10%

0% $20k $40k $60k $80k $100k Target Individual Household Income

perceived differences in spare money would in turn influence giving Table 2 standards. That is, holding income and expenditures constant, in- The effect of participant’s relative income on their judgments of howmucha dividuals who save with a specific goal in mind will be perceived as given target individual should donate to charity (as a percent of the target’s having less spare money, and less obligated to donate to charity than total household income) and how much spare money they expect the target those who save equivalent sums of money, but do not allocate this income to have. Relative income is measured in units of $10,000. money towards a savings goal. Subjective Giving Standards Estimated Spare Money

Target βRelativeIncome S.E. t βRelativeIncome S.E. t 7.1. Methods Income We recruited 298 participants (56% female; Mean Age = 35; $20,000 −0.39% 0.12 −3.22** −0.09 0.02 −4.41*** $40,000 −0.69% 0.12 −5.72*** −0.14 0.02 −6.66*** Median Household Income = $45,000 to $50,000 a year) via $60,000 −0.81% 0.12 −6.78*** −0.14 0.02 −6.67*** Mechanical Turk. Participants read about Mike and Mary, a couple $80,000 −1.04% 0.12 −8.64*** −0.09 0.02 −4.35*** whose household income is $50,000 a year, leaving them with $42,000 $100,000 −1.20% 0.12 −10.00*** −0.06 0.02 −2.72** after taxes. Participants were then presented with a table containing four rows describing how much the couple spends each year. The first two rows showed that the couple spends $14,000 on rent and $16,000 7. Study 3 on other expenses. In the Committed Savings condition, the latter two rows showed that the couple allocates $10,000 a year to their retire- The studies thus far show a concordance between perceptions of ment savings and have $2000 a year in uncommitted savings. In the spare money and subjective ethical standards of giving. The present Uncommitted Savings condition, these numbers were reversed, such study assesses whether the manner in which people allocate their in- that Mike and Mary were shown to allocate $2000 a year to their re- come casually affects perceived spare money and judged giving stan- tirement savings and have $10,000 a year in uncommitted savings. dards. One reason people overestimate the effects of earning a higher Participants then reported how much they believe that Mike and income is that they change the way they think about their finances as Mary should donate to charity each year in dollars. They then re- soon as they earmark money for a specific purpose (Thaler, 1985). sponded to two questions assessing their subjective perceptions of the Earmarking often acts as a pre-commitment device that individuals are couple’s spare money. This included “How much spare money do Mike reluctant to violate (Heath & Soll, 1996; Shefrin & Thaler, 1988), even and Mary have?” and “How financially constrained are Mike and when doing so could be financially advantageous (Sussman & O’Brien, Mary?” on seven-point scales. The latter measure was reverse coded and 2016). combined to create a two-item measure of perceived spare money In this study, we examine how earmarking affects perceived spare (r = 0.59, p < .001). money and subjective giving standards. We expect that putting aside money for a specific purpose will decrease perceptions of spare money even when that purpose is a long-term savings goal. In contrast, we 7.2. Results expect that money that is saved with no explicit purpose in mind will be encoded as spare money. Though in both cases the same amount of Due to the skewness in subjective giving standards responses, we money is being saved rather than spent, we expect that earmarking will utilize a Mann-Whitney U test to analyze the data. Results show that decrease subjective perceptions of spare money, compared to saving when Mike and Mary had committed less money to retirement savings, with no defined goal in mind. More importantly, we expect that any they were expected to donate more to charity (M = $1,171,

51 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56

SD = $1,717) than when they had committed more money to retire- affects their subjective judgments of giving standards (13 studies, ment savings (M = $866, SD = $1,809), U = 13,231.5, Z = 2.94, N = 6360) and judgments of spare money (15 studies, N = 7163). p = .004. We found the same pattern of results for the perceived spare Subjective giving standards were measured in dollar amounts in nine money measure. When Mike and Mary committed less money to re- studies and as a percentage of total income in four studies. Dollar tirement savings, participants thought they had more spare money amounts were log transformed (ln + 1) before conducting all analyses, (M = 4.23, SD = 1.33) than when they committed more to retirement and all results also hold with untransformed dollar amounts. savings (M = 3.52, SD = 1.41), U = 14,396, Z = 4.46, p < .001.2 To conduct the median target income analysis, we calculated a Mediation Analysis: We ran a bootstrapped mediation analysis with Pearson correlation between participant household income and the DV 10,000 samples to examine whether the effect of uncommitted savings of interest. To conduct the first response analysis, we calculated a on giving standards is mediated by estimated spare money. This med- standardized regression coefficient between relative income and theDV iation model used logged giving standards as the DV, savings condition of interest, controlling for target income level (Peterson & Brown, as the IV, and the two-item measure of perceived spare money as the 2005). We computed Fisher’s Z for each correlation from each study mediator variable. Consistent with our expectations, we find a sig- and calculated a weighted mean Z across studies, using inverse variance nificant indirect effect, as indicated by a confidence interval thatex- weights to assign more weight to studies with larger samples. We cludes zero (β = 0.35, SE = 0.11, 95% CI [0.18, 0.60]). When Mike and converted these Fisher’s Z values back to r for ease of presentation and Mary had less money committed to retirement savings, participants interpretation (Lipsey & Wilson, 2001). Given the similarities in design perceived them to have more spare money than when they had a higher and samples across studies, we present fixed effects models in the commitment to retirement savings (a = 0.71, S.E. = 0.16, p < .001), and greater estimates of spare money predicted higher giving standards (b = 0.49, S.E. = 0.09, p < .001). Controlling for the mediator, the effect of relative income on giving standards decreased from c=0.49, SE = 0.25, p = .05 to c’ = 0.14, SE = 0.25, p = .57.

7.3. Discussion

This study shows how the manner in which people budget their money casually affects subjective ethical standards of giving. Even holding income and expenditures constant, those who save for a specific long-term goal are seen as having less spare money, and therebyas less obligated to donate to charity, than those who save the same amount without any explicit goal.

8. Meta-analysis

In the preceding sections, we presented only 4 out of the 18 studies we conducted for this project. In order to assess the robustness of our findings, we conducted a meta-analysis of our entire file-drawer.

8.1. Methods & results

Of the 18 studies conducted, our meta-analysis includes the 15 studies that directly assess judged giving standards and spare money. Specifically, we eliminate three vignette studies that provided explicit information about the spare money of hypothetical individuals instead of measuring participants’ subjective perceptions of spare money. The studies included in the meta-analysis were designed to examine the robustness of the basic effect to various factors (e.g., order effects, utilizing different response scales), and we present an overview ofeach of these studies in the supporting materials. Of the 15 included studies, six asked participants to make a judgment of a single target individual, while the other nine asked participants to make judgments of multiple target individuals. For the studies that included multiple target individuals, we utilized two different se- lection criteria to enable us to make consistent comparisons with the results of our single-target studies: (1) a “Median Target Income” analysis that evaluates only responses towards the target individual that earns an income closest to $55,000 (the median household income in the United States at the time of the studies); and (2) a “First Response” analysis that evaluates only the first assessment made by participants, controlling for target income level when applicable. We examined the extent to which a participants’ relative income Fig. 4. a & b. Forest plots examining the total effect of relative income on 2 These results hold when a t-test is run on the log transformed subjective subjective giving standards (top panel) and perceived spare money (bottom ethical standards DV, t(2 9 6) = 1.96, p = .05, d = 0.23, and the un- panel) for the median target income analysis. For details on corresponding transformed two-item spare money DV, t(2 9 6) = 4.49, p < .001, d = 0.52. studies see supporting materials.

52 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56

analyses below. For additional robustness checks, including random In this regression, β1 evaluates the magnitude of the relative income effects models, see our supporting materials. effect for individuals making upward evaluations (towards higher-

As predicted, the relationship between relative income and sub- earning others) while β2 evaluates the magnitude of the relative income jective giving standards is significant and negative for both the median effect for individuals making downward evaluations (towards lower- target income analysis (r = −0.13, 95% CI = [−0.16, −0.11], earning others). For the first response analysis, we utilize the equation Z = −10.67, p < .001; see Fig. 4a) and the first response analysis above, and further included a control variable for target income. For (r = −0.13, 95% CI = [−0.16, −0.11], Z = −10.58, p < .001). Ad- each regression, we then extracted the standardized correlation be- ditionally, the relationship between relative income and perceptions of tween the coefficient and the DV(Peterson & Brown, 2005). We then spare money is also significant and negative for both the median target converted this value to Fisher’s Z using the same procedure described income analysis (r = −0.29, 95% CI = [−0.32, −0.27], Z = -24.98, above. p < .001; see Fig. 4b) and the first response analysis (r = −0.23, 95% Subjective Giving Standards: Results suggest that the effect of relative CI = [−0.26, −0.21], Z = -19.95, p < .001). Thus, the full set of income is stronger for upward evaluations of higher-earning others than studies provides consistent evidence that a participant’s income relative for downward evaluations of lower-earning others. For the median in- to a target affects their perceptions of that target’s spare money and the come analysis, we find that when making upward evaluations, the less standards of giving they apply to them. that participants earn, the more they believe that the target individual should donate to charity, R = -0.11, 95% CI = [-0.14, -0.09], Z = - 8.2. Asymmetric sensitivity to upward vs. Downward evaluations 8.94, p < .001. However, for those who earn more than the target, thereby making downward evaluations, this relationship is notably In the analyses presented throughout the paper, we have shown that weaker, R = -0.04, 95% CI = [-0.06, -0.01], Z = -3.03, p < .001. estimates of spare money and subjective giving standards are reference- Similarly, in the first response analysis, the relationship between re- dependent with respect to the self. In the present section, we built on lative income and subjective giving standards is notably stronger for our prior analyses by investigating whether individuals show differ- upward evaluations of higher earners, R = -0.10, 95% CI = [-0.13, ential sensitivity to changes in income when making upward evalua- -0.08], Z = -8.06, p < .001, than for downward evaluations of lower tions of higher earners (i.e., targets who earn more than the participant) earners, R = -0.06, 95% CI = [-0.08, -0.03], Z = -4.69, p < .001. versus downward evaluations of lower earners (i.e., targets who earn To test whether the difference in slopes is significant, we conducted less than the participant). On the one hand, consistent with general a model comparison test comparing Eq. (2), which estimates the slope negativity bias (Rozin & Royzman, 2001) and loss aversion (Kahneman of the effect separately for those who earn more versus those whoearn & Tversky, 1979), people may be particularly sensitive to losses in in- less than the target income, to a model that does not differentiate be- come and thus perceive lower earners as especially constrained for tween those who earn more or less than the target (Giving Stan- spare money. dards = α + β1(RelativeIncome) + ε). We calculated the semi-partial r On the other hand, if people already feel constrained for spare of the difference between these two models, and assigned a valence to money across most income levels, there may be little room to adjust the results depending on the direction of the effect, such that a positive their judgments of lower-earning individuals further downwards. sign indicates a stronger effect of relative income when making upward Conversely, judgments of higher-earning individuals are not bounded in evaluations of higher earners, while a negative sign indicates a stronger this way. Hence, people may expect that spare money increases with effect of relative income when making downward evaluations oflower income gains more than it decreases with income losses, since spare earners. Results show that the model comparison across studies is sig- money can only fall so low before approaching zero. The same rea- nificant and positive for the median target income analysis, soning applies to judgments of giving standards. If people across income RModelComparison = 0.04, 95% CI = [0.02, 0.07], Z = 3.35, p < .001, levels feel obligated to donate very little, then the standards they assign and marginally significant for the first response analysis, to poorer others may have little room to further decrease. However, if RModelComparison = 0.02, 95% CI = [0.00, 0.05], Z = 1.87, p = .06. greater income is expected to significantly increase spare money, then These results indicate an asymmetric effect around the reference point people may believe that higher earners are obligated to donate sig- of one’s own income, whereby participants are more responsive to gains nificantly more to charity. in income than losses in income relative to the self when evaluating To test these possibilities, we use a piecewise regression model to ethically appropriate levels of charitable giving. estimate the relationship between relative income and our focal DVs Spare Money: We repeated the above analysis for estimated spare (Subjective Giving Standards and Expected Spare Money) separately for money DV and again find evidence that participants are more re- those who earn less than the target (and thereby make upward eva- sponsive to upward evaluations of higher earners compared to down- luations), and those who earn more than the target (and thereby make ward evaluations of lower earners. For the median income analysis, downward evaluations). The model fits two lines that are constrained to when isolating participants making upward evaluations of higher meet at the target income level, and thus allows us to statistically test earners, we find that the less that participants earn relative to the target for potential differences in the effect of relative income on upwards individual, the more spare money they believe that the target individual versus downwards evaluations. We again conducted both a median has, R = -0.19, 95% CI = [−0.21, −0.17], Z = −16.11, p < .001. target income analysis and a first response analysis. For the median This relationship is weaker when isolating participants making down- target income analysis, we employed the following regression for each ward estimations of lower-earning targets, R = −0.14, 95% study: CI = [−0.17, −0.12], Z = −12.13, p < .001. Similarly, in the first response analysis, the relationship between relative income and spare Giving Standards money is again stronger for individuals making upward evaluations of = + 1 (RIUpwardEvaluation)+ 2 (RIDownwardEvaluation) + (2) higher earners, R = −0.16, 95% CI = [−0.19, −0.15], Z = −13.89, p < .001, than for individuals making downward estimations of lower RI_UpwardEvaluation= Income of participant to relative to target for all earners, R = −0.11, 95% CI = [−0.14, −0.09], Z = −9.56, values< 0, else = 0; p < .001. To test the relative effect, we repeated the model comparison pro- RI_DownwardEvaluation= Income of participant to relative to target for cedure, which yielded significant positive results for both the median all values> 0, else = 0. target income analysis, RModelComparison = 0.04, 95% CI = [0.02, 0.06], Z = 3.63, p < .001, and the first response analysis,

53 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56

RModelComparison = 0.05, 95% CI = [0.02, 0.07], Z = 3.87, p < .001. Office, 2011; Havens, O’Herlihy, & Schervish, 2006; James &Sharpe, These findings provide convergent evidence that relative income hasa 2007; Schervish & Havens, 2001). Despite this weak relationship, our stronger effect on upward evaluations of higher earners than downward data suggests that the relatively poor set much higher standards for how evaluations of lower earners. much higher earners should donate than the standards those higher earners set for themselves or similar others (even for incomes far below 9. General discussion the $350,000 threshold). While we have focused on subjective standards of charitable giving, This paper investigates how individuals derive charitable standards it is possible that similar effects hold for forecasts of other discretionary for the self and others, and presents three main conclusions. First, we expenditures. For instance, people might overestimate how many va- find that judgments regarding ethical standards of giving are reference- cations richer others can afford to take or how big a house theycan dependent with respect to one’s own income. Consistent with the afford to buy. However, it is less clear whether people also expect that principle of diminishing marginal utility for income, we find that richer others should spend more on such expenditures than those richer people expect that higher earners should give more than lower earners. others believe they should spend. To examine this possibility, we ran a However, we additionally find that these judgments are affected byhow study (supporting materials, Study S1, N = 500) in which participants much individuals themselves earn, which cannot be explained by di- evaluated how much someone who earns $50,000 should spend on minishing marginal utility alone. Thus, giving standards are determined clothing, transportation, healthcare, housing, and donations to charity jointly by the evaluator’s income and the target’s income, not just the each year. Only charitable donations were negatively associated with latter. participant income (β = −0.04, SE = 0.02, p = .018), such that the Second, we find that subjective standards of giving are affected by less participants earned, the more they believed that target should mispredictions regarding the relationship between income and spare donate to charity. In contrast, expenditures on clothing, housing, and money. People expect that higher-earning others have more spare healthcare were not significantly associated with participant income money than those higher-earners actually report having. Accordingly, (ps > 0.10), while transportation spending was positively associated we connect the financial decision-making literature to research on with participant income (β = 0.03, SE = 0.01, p = .036). ethical decision-making, and demonstrate that misperceptions of fi- Even other expenditures with a strong normative component, such nancial constraints shape the generosity standards that people apply to as expected tax obligations, may not show these effects as strongly or themselves and others. consistently as giving obligations. An additional study (supporting Third, a meta-analysis provides initial evidence of a difference in materials, Study S2, N = 501), found no significant relationship be- how individuals estimate the spare money of those who earn more than tween participant income and the amount they believe someone them versus those who earn less than them. People significantly over- making $30,000 should be taxed (β = 0.02, SE = 0.12, p = .89) or estimate the spare money of higher earners, but evaluate the spare $65,000 (β = −0.16, SE = 0.12, p = .18), but did find that the less money of lower earners with greater accuracy. Subjective giving stan- participants earned, the higher they believed the tax rate should be for dards follow the same pattern: lower earners believe that higher earners an individual earning $100,000 (β = −0.31, SE = 0.12, p = .01).3 The ought to give much more to charity than those higher earners believe discretionary nature of charitable giving may make these judgments they ought to give. However, higher earners and lower earners show particularly sensitive to relative income effects, compared to tax ob- greater agreement in the charitable standards they assign to lower ligations, which people often find distasteful (Sussman & Olivola, 2011) earners. and are required by law to fulfill. Future research could further explore If individuals overestimate the relationship between gains in income how different sorts of financial obligations are thought to change with and spare money, they may feel justified in forgoing donation until income. their income increases (see Zauberman & Lynch, 2005, Study 7). However, our results support the notion that any gains in perceptions of 9.2. Barriers to establishing standards for giving spare money are short-lived. Once people do earn more money, increase their expenditures, and again find themselves strapped for cash, they In our studies, we asked participants to evaluate what amount of may put off donating until they earn even more. To the extent that money would be morally appropriate to donate to charity. However, it individuals believe that their own idiosyncratic financial constraints are is less clear whether (or what proportion of) individuals feel that giving more exceptional than others (cf. Davidai & Gilovich, 2016), they may represents a duty which individuals are obligated to fulfill or whether feel even more justified in donating very little in the present. Ifso, giving is considered supererogatory (i.e., commendable, yet optional). individuals may maintain the belief that they are generous without Some individuals certainly do feel a moral duty to donate. For in- actually contributing much to charity, continually “passing the buck” to stance, religious institutions often advocate or enforce “tithing,” in wealthier others or to their future, wealthier selves. Given this con- which followers are expected to donate a minimal portion of their in- tinual temporal shift in subjective standards of giving, commitment come to charity, typically 10%. Some philosophers argue that donating devices may be useful in combating this tendency by enabling people to to charity ought to be a moral obligation, even if many people do not pre-commit future increases in spare money to charity in advance of a view it accordingly. In particular, the effective altruism movement was raise (Breman, 2011). born from consequentialist ethical arguments that donating to save the lives of others is just as obligatory as saving the life of a drowning child 9.1. An “Expectations Gap” in charitable giving (Singer, 2009, 2015; MacAskill, 2015). Even for those who agree with the notion that people ought to be More broadly, our research indicates the existence of an expecta- duty-bound to donate, establishing clear standards for giving remains tions gap between how much people believe that richer others should elusive. For instance, a flat giving rate applied to a percentage ofone’s donate to charity and how much those richer others actually donate. income, such as a tithe, may be seen as overly burdensome to the Investigations into the actual relationship between income and chari- poorest members of society and thus seem unappealing. For instance, in table giving show a weak relationship at best. Evaluations of tax data in Study 2, only 18% of participants applied a constant donation the United States shows that the proportion of income donated to charity remains constant across most income levels, except for particularly high-earning individuals (those who make more than approxi- 3 See also supporting materials, Study S3 which finds a non-significant effect mately $350,000 a year), who donate a greater proportion of their in- for perceptions of how much someone who earns $50,000 should be taxed (β= come to charity than lower-earning Americans (Congressional Budget −0.36, SE = 0.24), t(3 0 1) = −1.52, p = .13).

54 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56 obligation across target incomes. In contrast, 71% of participants in- 9.4. Spare resources, the hedonic treadmill, and subjective well-being dicated that donation obligations ought to be progressive, such that higher earners should be expected to donate a greater percentage than The research presented here dovetails with research on the “hedonic lower earners. treadmill”, or the notion that emotional experience are highly sensitive Moreover, even if most people believe that donation rates as a to changes in states, but that people ultimately adapt to those changes percentage of income should be progressive, our results indicate sys- over time (Brickman & Campbell, 1971; Diener, Lucas & Scollon, 2006). tematic heterogeneity in beliefs about how much individuals with dif- The hedonic treadmill has been used to explain why income increases ferent incomes can and should give. Such heterogeneity might preclude do not ultimately lead to significant change in subjective well-being: any clear consensus on what donation standards would most benefit even if people obtain a momentarily boost in pleasure from receiving a society. Future research might shed further light on additional barriers raise, they eventually adapt and return close to their previous baseline to forming clear giving standards and the downstream consequences of feelings (Frederick & Loewenstein, 1995; Kahneman et al., 2006). attempting to establish stronger norms around appropriate levels of While the hedonic treadmill predicts that people will adapt to giving. For instance, establishing stronger social norms might reduce changing levels of wealth, it remains unclear how adaptation affects the individuals’ sense of agency over their donation decisions, thus redu- way people manage their finances. However, we find that just as in- cing the enjoyment they derive from giving (Harbaugh, Mayr, & dividuals erroneously predict the extent to which earning more money Burghart, 2007; Weinstein & Ryan, 2010; Berman & Small, 2012). If so, will make them happier, they also erroneously predict the extent to then strengthening expectations regarding how much people ought to which it will increase the spare money available to them. It is likely that donate may be met with resistance, even from committed donors. these effects interact. For instance, higher absolute income levels and Even if establishing prescriptive or injunctive norms remains elu- greater reserves of spare money are positively associated with sub- sive, research suggests that making good deeds public can establish jective well-being, presumably because they eliminate stress associated stronger descriptive norms around generosity, thus increasing donation with making ends meet (Ruberton, Gladstone & Lyubomirsky, 2016; rates (Croson, Handy, & Shang, 2009; Goldstein, Cialdini, & Sacks, Stevenson, and Wolfers, 2012). Future research can investigate Griskevicius, 2008; Kraft-Todd, Yoeli, Bhanot, & Rand, 2015). Yet, we the extent to which individuals can translate income gains to long know little about how publicizing the good deeds of the wealthy might lasting increases in subjective well-being. influence the behavior of the poor, or vice versa. The present research suggests that high giving rates among the wealthy may not necessarily 10. Conclusion inspire others to donate. We find that as the difference between an evaluator and a target increases, individuals’ beliefs about their own In conclusion, this paper finds evidence that subjective standards of subjective generosity standards are contrasted with those of higher- giving are reference dependent, and affected by misperceptions of the earners (cf. Mussweiler, 2003). Publicizing the philanthropic pursuits of relationship between income and spare money. Individuals feel the wealthy may also make lower earners feel relatively incapable of strapped for cash and thereby hold themselves to a relatively low making an impact. For instance, the Giving Pledge, a campaign pro- standard of giving, but believe that higher-earning others (and their moted by Bill Gates and Warren Buffett to increase the rate of giving higher-earning future selves) are flush with spare money that makes it among billionaires has seen promises of over $500 billion to charity. easy for them to give. In contrast, those higher earners also feel Relative to these massive sums, a single donation by a less wealthy strapped for cash and hold themselves to lower standards, thus giving donor may feel like a drop in the bucket. Future research can examine far less than poorer others feel they ought to give. Regardless of their under what conditions actors are influenced by the good deeds of others current income, people seem to perceive their current selves as un- whose resources and constraints differ from their own. iquely constrained, and thus free from responsibilities to be generous to others.

9.3. Resource contributions in organizations Appendix A. Supplementary material

Although our studies focus on individual charitable donations, the Supplementary data to this article can be found online at https:// effects we uncover likely apply towards resource contributions insocial doi.org/10.1016/j.obhdp.2019.12.005. collectives and other organizational units. Small units within an orga- nization may expect that larger units with greater resources or prowess References have enough spare capacity to take on additional burdens. However, the present research suggests that those larger units may view their own Bekkers, R., & Wiepking, P. (2011). A literature review of empirical studies of philan- capacity as much more constrained than those smaller units expect. thropy: Eight mechanisms that drive charitable giving. Nonprofit and Voluntary Sector Quarterly, 40(5), 924–973. Even in the absence of objective differences in available resources, Berman, J. Z., & Small, D. A. (2012). Self-interest without selfishness: The hedonic benefit individuals or organizational units may tend to see themselves as par- of imposed self-interest. Psychological Science, 23(10), 1193–1199. ticularly strapped for spare resources compared to similar others. Berman, J. Z., Tran, A. T., Lynch, J. G., Jr, & Zauberman, G. (2016). Expense neglect in forecasting personal finances. Journal of Marketing Research, 53(4), 535–550. Research shows that people tend to be more sensitive to the barriers Breman, A. (2011). Give more tomorrow: Two field experiments on altruism and inter- they have faced than the blessings they have received, and furthermore temporal choice. Journal of Public Economics, 95(11), 1349–1357. believe that they have confronted greater barriers than others have. For Brickman, P., & Campbell, D. T. (1971). Hedonic relativism and planning the good so- example, both Democrats and Republicans are more likely to believe ciety. In M. H. Appley (Ed.). Adaptation-level theory (pp. 287–305). . Congressional Budget Office (2011). Options for changing thetax treatment of charitable that the Electoral College harms their party more than it helps them, giving. CBO Report, Congressional Budget Office, Washington D.C. and individuals are more likely to feel that their parents were harsher Croson, R., Handy, F., & Shang, J. (2009). Keeping up with the Joneses: The relationship on them than on their siblings (Davidai & Gilovich, 2016). Just as of perceived descriptive social norms, social information, and charitable giving. Nonprofit Management and Leadership, 19(4), 467–489. people believe their contributions to joint efforts are greater than others Davidai, S., & Gilovich, T. (2016). The headwinds/tailwinds asymmetry: An availability (Ross & Sicoly, 1979), individuals may feel that their personal sacrifices bias in assessments of barriers and blessings. Journal of Personality and Social in support of organizations are more burdensome than those made by Psychology, 111(6), 835–851. Diener, E., Lucas, R. E., & Scollon, C. N. (2009). Beyond the hedonic treadmill: Revising the others. When asked to “go above and beyond,” people may believe that adaptation theory of well-being. In The science of well-being. Dordrecht: overcoming their idiosyncratic resource constraints is particularly on- Springer103–118. erous, without realizing that others feel similarly about their own re- Frederick, S., & Loewenstein, G. (1999). Hedonic adaptation. In D. Kahneman, & E. Diener (Eds.). Well-being: The Foundations of Hedonic Psychology (pp. 302–329). New York, source constraints.

55 J.Z. Berman, et al. Organizational Behavior and Human Decision Processes 157 (2020) 46–56

NY: Russell Sage. net worth philanthropy. Bloomington, IN: The Center on Philanthropy at Indiana Frederick, S., Novemsky, N., Wang, J., Dhar, R., & Nowlis, S. (2009). Opportunity cost University. neglect. Journal of Consumer Research, 36(4), 553–561. Ross, M., & Sicoly, F. (1979). Egocentric biases in availability and attribution. Journal of Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint: Using Personality and Social Psychology, 37(3), 322. social norms to motivate environmental conservation in hotels. Journal of Consumer Rozin, P., & Royzman, E. B. (2001). Negativity bias, negativity dominance, and contagion. Research, 35(3), 472–482. Personality and Social Psychology Review, 5(4), 296–320. Harbaugh, W. T., Mayr, U., & Burghart, D. R. (2007). Neural responses to taxation and Ruberton, P., Gladstone, J., & Lyubomirsky, S. (2016). How Your Bank Balance Buys voluntary giving reveal motives for charitable donations. Science, 316(5831), Happiness. Emotion, 16(5), 575–580. 1622–1625. Sacks, D. W., Stevenson, B., & Wolfers, J. (2012). The new stylized facts about income and Havens, J. J., O’Herlihy, M. A., & Schervish, P. G. (2006). Charitable giving: How much, subjective well-being. Emotion, 12(6). by whom, to what, and how. The Nonprofit Sector: A Research Handbook, 2, 542–567. Schervish, P. G., & Havens, J. J. (2001). Wealth and the commonwealth: New findings on Heath, Chip, & Soll, Jack B. (1996). Mental budgeting and consumer decisions. Journal of wherewithal and philanthropy. Nonprofit and Voluntary Sector Quarterly, 30(1), 5–25. Consumer Research, 23(June), 40–52. Sharma, E., Mazar, N., Alter, A. L., & Ariely, D. (2014). Financial deprivation selectively Kahneman, D., Krueger, A. B., Schkade, D., Schwarz, N., & Stone, A. A. (2006). Would you shifts moral standards and compromises moral decisions. Organizational Behavior and be happier if you were richer? A focusing illusion. Science, 312(5782), 1908–1910. Human Decision Processes, 123, 90–100. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Shefrin, Hersh M., & Thaler, Richard H. (1988). The behavioral life-cycle hypothesis. Econometrica: Journal of the Econometric Society, 263–291. Economic Inquiry, 26(4), 609–643. James, R. N., III, & Sharpe, D. L. (2007). The nature and causes of the U-shaped charitable Singer, P. (1979). Practical ethics. Cambridge: Cambridge University Press. giving profile. Nonprofit and Voluntary Sector Quarterly, 36(2), 218–238. Singer, P. (2009). The life you can save: Acting now to end world poverty. New York: Random Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The endowment effect, House. loss aversion, and status quo bias. The Journal of Economic Perspectives, 5(1), 193–206. Singer, P. (2015). The most good you can do: How effective altruism is changing ideas about Kraft-Todd, G., Yoeli, E., Bhanot, S., & Rand, D. (2015). Promoting cooperation in the living ethically. New Haven: Yale University Press. field. Current Opinion in Behavioral Sciences, 3, 96–101. Slovic, P. (2007). ‘‘If I look at the mass I will never act”: Psychic numbing and genocide. LaBarge, M. C., & Stinson, J. L. (2014). The role of mental budgeting in philanthropic Judgment and Decision Making, 2(2), 79–95. decision-making. Nonprofit and Voluntary Sector Quarterly, 43(6), 993–1013. Small, D. A., & Cryder, C. (2016). Prosocial consumer behavior. Current Opinion in Lipsey, M.W. & Wilson, D. B. (2001). Practical Meta-Analysis (Vol. 49). Thousand Oaks, Psychology, 10, 107–111. CA: Sage publications. Sussman, A. B., & Alter, A. L. (2012). The exception is the rule: Underestimating and Loewenstein, G. (1996). Out of control: Visceral influences on behavior. Organizational overspending on exceptional expenses. Journal of Consumer Research, 39(4), 800–814. Behavior and Human Decision Processes, 65(3), 272–292. Sussman, A. B., & O’Brien, R. L. (2016). Knowing when to spend: Unintended financial Lusardi, A., Schneider, D. J., & Tufano, P. (2011). Financially fragile households: consequences of earmarking to encourage savings. Journal of Marketing Research, Evidence and implications (No. w17072). National Bureau of Economic Research. 53(5), 790–803. Lynch, J. G., Jr, Netemeyer, R. G., Spiller, S. A., & Zammit, A. (2009). A generalizable Sussman, A. B., Sharma, E., & Alter, A. L. (2015). Framing charitable donations as ex- scale of propensity to plan: The long and the short of planning for time and for ceptional expenses increases giving. Journal of Experimental Psychology: Applied, money. Journal of Consumer Research, 37(1), 108–128. 21(2), 130–140. MacAskill, W. (2015). Doing Good Better: Effective Altruism and a Radical New Way toMake Sussman, A. B., & Olivola, C. Y. (2011). Axe the tax: Taxes are disliked more than a Difference. London: Guardian Faber Publishing. equivalent costs. Journal of Marketing Research, 48(SPL), S91–S101. Mussweiler, T. (2003). Comparison processes in social judgment: Mechanisms and con- Thaler, R. (1985). Mental accounting and consumer choice. Marketing Science, 4(3), sequences. Psychological Review, 110(3), 472. 199–214. Peetz, J., & Buehler, R. (2009). Is there a budget fallacy? The role of savings goals in the UBS Investor Watch (2013), “What Is ‘Wealthy’?” https://www.ubs.com/content/dam/ prediction of personal spending. Personality and Social Psychology Bulletin, 35(12), WealthManagementAmericas/documents/investor-watch-3Q2013-report.pdf. 1579–1591. Weinstein, N., & Ryan, R. M. (2010). When Helping Helps: Autonomous motivation for Peterson, R. A., & Brown, S. P. (2005). On the use of beta coefficients in meta-analysis. prosocial behavior and its influence on well-being for the helper and recipient. Journal of Applied Psychology, 90(1). Journal of Personality and Social Psychology, 98(2), 222–244. Piff, P. K., Kraus, M. W., Côté, S., Cheng, B. H., & Keltner, D. (2010). Having less,giving Zauberman, G., & Lynch, J. G., Jr (2005). Resource slack and propensity to discount more: The influence of social class on prosocial behavior. Journal of Personality and delayed investments of time versus money. Journal of Experimental Psychology: Social Psychology, 99(5), 771–784. General, 134(1), 23–37. Rooney, P. M., & Frederick, H. K. (2007). Portraits of donors: Bank of America study of high